Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Conditional input in verification property #785

Closed
AWbosman opened this issue Mar 26, 2024 · 4 comments
Closed

Conditional input in verification property #785

AWbosman opened this issue Mar 26, 2024 · 4 comments

Comments

@AWbosman
Copy link

Dear authors,

I am interested in verifying a property where the change a certain input variable can make is dependent on another.
In a comment on an issue @MatthewDaggitt gave an example ( x0 +x1 <=1) and I was wondering whether this kind of property is currently supported by Marabou as I have not seen any examples of this.

Originally posted by @MatthewDaggitt in #490 (comment)

@wu-haoze
Copy link
Collaborator

wu-haoze commented Mar 26, 2024

Hi @AWbosman , Marabou supports general linear constraints over input variables. I usually prefer using the Python API: https://github.com/NeuralNetworkVerification/Marabou/blob/master/maraboupy/examples/7_PythonicAPI.py

Other front ends of Marabou (e.g., VNN_LIB parser) can also take constraints like that.

@AWbosman
Copy link
Author

AWbosman commented Mar 26, 2024

Hi wu-haoze,

Thanks for the quick reply!
I will close the issue

EDIT: Could you direct me to an explanation on how to use a vnnlib file as veriifcation property with Marabou? I have a vnnlib as well as a network.

I have tried running command line like this : Marabou opset_10_nopad.onnx spec.vnnlib
There are no errors, but it runs forever and I cannot seem to set a timeout on commandline.

@wu-haoze
Copy link
Collaborator

wu-haoze commented Apr 1, 2024

@AWbosman , this is the right way to use it on the command line. ./Marabou --help will show the command line options, including timeout.

The long running time could be due to the sub-optimal implementation of the C++ onnx parser (#754 ) which we are trying to fix. For now, please use the Python API if you want to parse a ONNX network.

@wu-haoze
Copy link
Collaborator

Closing the issue due to inactivity. Feel free to re-open if needed.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants